mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-12-30 21:34:36 +00:00
2bf8d0f7c4
* backend : offload large batches to GPU * fix hip * code cleanup * fix CUDA split buffers * Update ggml-backend-impl.h Co-authored-by: Johannes Gäßler <johannesg@5d6.de> * cuda : fix memset without set_device * imatrix : remove sched affix from weight names * sched : add a new split if the current one has too many inputs reduce max inputs per split more cleanup * update backends ggml-ci --------- Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
44 lines
1.4 KiB
C
44 lines
1.4 KiB
C
#pragma once
|
|
|
|
#include "ggml.h"
|
|
#include "ggml-backend.h"
|
|
|
|
#ifdef GGML_USE_HIPBLAS
|
|
#define GGML_CUDA_NAME "ROCm"
|
|
#define GGML_CUBLAS_NAME "hipBLAS"
|
|
#else
|
|
#define GGML_CUDA_NAME "CUDA"
|
|
#define GGML_CUBLAS_NAME "cuBLAS"
|
|
#endif
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
#define GGML_CUDA_MAX_DEVICES 16
|
|
|
|
// backend API
|
|
GGML_API GGML_CALL ggml_backend_t ggml_backend_cuda_init(int device);
|
|
|
|
GGML_API GGML_CALL bool ggml_backend_is_cuda(ggml_backend_t backend);
|
|
|
|
// device buffer
|
|
GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_buffer_type(int device);
|
|
|
|
// split tensor buffer that splits matrices by rows across multiple devices
|
|
GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_split_buffer_type(const float * tensor_split);
|
|
|
|
// pinned host buffer for use with the CPU backend for faster copies between CPU and GPU
|
|
GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_cuda_host_buffer_type(void);
|
|
|
|
GGML_API GGML_CALL int ggml_backend_cuda_get_device_count(void);
|
|
GGML_API GGML_CALL void ggml_backend_cuda_get_device_description(int device, char * description, size_t description_size);
|
|
GGML_API GGML_CALL void ggml_backend_cuda_get_device_memory(int device, size_t * free, size_t * total);
|
|
|
|
GGML_API GGML_CALL bool ggml_backend_cuda_register_host_buffer(void * buffer, size_t size);
|
|
GGML_API GGML_CALL void ggml_backend_cuda_unregister_host_buffer(void * buffer);
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|